|Publication number||US8023441 B2|
|Application number||US 11/312,269|
|Publication date||Sep 20, 2011|
|Filing date||Dec 20, 2005|
|Priority date||Dec 20, 2004|
|Also published as||US8792402, US20060187866, US20110299421, WO2006069067A2, WO2006069067A3|
|Publication number||11312269, 312269, US 8023441 B2, US 8023441B2, US-B2-8023441, US8023441 B2, US8023441B2|
|Inventors||Jay Philip Werb, Victor A. Berry|
|Original Assignee||Sensicast Systems|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (39), Non-Patent Citations (1), Referenced by (6), Classifications (27), Legal Events (7)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This application claims the benefit of U.S. Provisional Application No. 60/637,644, filed Dec. 20, 2004, which is hereby incorporated by reference in its entirety.
1. Field of Invention
This invention relates to wireless communication networks in which wireless communication devices send information to each other and/or a host computer.
2. Description of Related Art
As advances in technology enable the development of ever-smaller wireless devices such as sensors and actuators, there has been increasing interest in self-configuring multihop wireless networks of these devices, together with additional communication devices and software. Such networks, typically known as Wireless Sensor Networks (WSNs), have a number of potential uses. For example, WSNs may be employed for automated meter reading (AMR) applications, such as for metering residential heat, electricity, or water usage.
In examining the operation of WSNs, some questions arise: How might data be reported from device to device within a WSN? How may data from a network of devices be accumulated and passed to devices or applications outside the WSN? What are some requirements for timely, low power communication within such WSNs? What features may be incorporated to improve system reliability? Various aspects of the invention described below relate to reporting and accumulating data in such wireless networks.
In one aspect of the invention, a method for coordinating communication in a wireless sensor network includes associating a first node with a wireless sensor network having at least two nodes associated with each other in a parent/child relationship. The first node may be associated with the wireless sensor network such that the first node functions as a parent or a child to at least one other node. A wake up time may be determined for the first node to wake up and listen for, or to broadcast, a signal regarding communication with a parent or child node. The wake up time may be determined based on a detected temperature at the first node. For example, operating temperatures local to a device, such as a router, edge node or other in a wireless sensor network, may affect the operation of a timer. Thus, differences in temperatures between different nodes in the network may affect the devices' ability to coordinate communications with each other if the communication is scheduled to occur at specified intervals. By adjusting for temperature conditions at one or more nodes, time periods in different devices may be more accurately or consistently measured. Any adjustment made for temperature may cause an adjustment to an elapsed time measurement made by a clock (e.g., an elapsed time period measured by a clock may be adjusted by some temperature varying factor) or to a expected time period (e.g., a wake up time period between communication intervals may be adjusted based on measure temperature to account for variations in clock operation).
In another aspect of the invention, a method for coordinating communication in a wireless sensor network includes associating a first node with a wireless sensor network having at least two nodes associated with each other in a parent/child relationship. The first node may be associated with the wireless sensor network such that the first node functions as a parent or child to at least one other node. A wake up time may be determined for the first node to wake up and listen for a signal regarding communication with a second node that has a parent or child relationship with the first node. The wake up time may be determined based on a determined clock skew between a clock in the first node and a clock in the second node. Manufacturing or other variations between clocks used in two or more devices in the network may result in time being measured differently by the devices. A node, such as an edge node, router or other, may determine a clock skew (a difference in time measurement) between the node and another device with which the node communicates. For example, an edge node may expect a parent router to beacon at a predetermined time, but in fact, the router may beacon at the predetermined time+/−a clock skew. For future communication times, the edge node may use the clock skew to adjust an elapsed time period or expected time period to determine when the router will wake up for communication.
In one aspect of the invention, a node may use temperature adjustment and a clock skew to determine a wake up time for communication with another node in a wireless sensor network. For example, a node may adjust elapsed time measured by its clock based on a measured temperature, and adjust wake up times when the node is scheduled to communication with one or more other nodes based on a clock skew determined for each of the other nodes. The node may use a different clock skew for each other node with which the node communicates so that differences in clock operation may be optimally compensated.
In another aspect of the invention, a method for forming a wireless sensor network may include providing a plurality of nodes at a location where at least two of the nodes are adapted and positioned relative to each other to form a wireless sensor network. A signal may be beaconed from a first node that represents that the first node is not associated with a wireless sensor network and that the first node is adapted to function as a parent node. A second node may be associated with the first node such that the second node is a child of the first node and the first node is a parent of the second node, thereby forming, at least in part, a wireless sensor network. A third node that is capable of functioning as a cluster head may initially be associated with the wireless sensor network as a child of an association node in the wireless sensor network, and thereafter the association node may be commanded to function as a child of the third node such that the third node functions as a parent to the association node. In accordance with this aspect of the invention, a cluster head, such as a data accumulator or gateway, may be associated with an existing wireless sensor network such that the cluster head is initially made a child of at least one node (e.g., a router) in the network. Thereafter, the cluster head may instruct one or more nodes in the network (including its initial one or more parents) to adjust their hierarchical function in the network so that the cluster head functions as a parent to nodes in the network. The cluster head may further adjust the hierarchical function of nodes in the network, such as reversing existing parent/child relationships. (Designation of “association node” does not necessarily require the node to have any specialized capability. Rather, an “association node” may be any device with which the cluster head initially associates. For example, the “first” or “second” node may function as the “association node.”)
In another aspect of the invention, a wireless sensor network includes a cluster head, a shadow cluster head, and a plurality of nodes in communication with the cluster head and the shadow cluster head. The plurality of nodes may include at least one router or at least one edge node, e.g., at least one of the nodes including an associated sensor to collect sensor data. The cluster head and the shadow cluster head may be adapted to receive communications from a same set of nodes, but the cluster head and the shadow cluster head may use different multipath profiles to communicate with nodes in radio range. For example, the cluster head may communicate wirelessly with a same set of nodes, but communicate with those nodes using different communication paths. Thus, the shadow cluster head may provide a redundant link from the set of nodes to the cluster head.
In another aspect of the invention, a wireless sensor network may include a data accumulator, and a plurality of nodes in communication with the cluster head. The plurality of nodes may include at least one router or at least one edge node that provides sensor data to the data accumulator, and the data accumulator may be adapted to communicate wirelessly with a mobile device to transmit the sensor data to the mobile device. The data accumulator and the mobile device may use a same radio MAC and PHY, and the mobile device may use an IEEE 802.15.4 radio to communicate with the data accumulator. The nodes and the data accumulator may communicate wirelessly at a frequency of about 868 MHz, and the data accumulator and the mobile device may communicate at a frequency of about 2.4 GHz.
In another aspect of the invention, a method for forming a wireless sensor network includes providing a plurality of routers that are configured and arranged to form a wireless sensor network in which routers have a hierarchical parent/child relationship. A gateway may be provided in communication with the wireless sensor network, where the gateway is adapted to send and receive signals for the wireless sensor network to at least one device outside of the wireless sensor network. The routers in the wireless sensor network may each have a hop count that indicates a number of routers through which communications from the router pass to the gateway, and routers with a lower hop count may have a larger battery capacity than routers with a higher hop count.
These and other aspects of the invention will be apparent from the following description and claims.
Aspects of the invention are described below with reference to the following drawings in which like numerals reference like elements, and wherein:
Aspects of the invention are described below with reference to illustrative embodiments. However, it should be understood that aspects of the invention are not limited to those embodiments described below, but instead may be used in any suitable system or arrangement.
Aspects of the invention are described in relation to a wireless sensor communication network (WSN) 6, in which routing and non-routing devices may be combined into a mesh or hierarchical structure, as shown, for example, in the illustrative embodiment of
As shown in
A WSN 6 may be designed to support hundreds of devices reporting daily. The WSN 6 may support only Routers 2 or permutations of Routers 2 and Edge Nodes 1 as appropriate for different applications. System capacity may primarily be limited by the battery life of Routers 2 close to Data Accumulators 3, as such Routers 2 may need to expend power passing traffic to and from a large number of more distant devices. System capacity may be increased with larger batteries on such Routers 2.
In this specification, “node” or “device” may refer to an Edge Node 1, Router 2, or other networked device. Within the system, Routers 2 and Edge Nodes 1 may have parent-child relationships, with Edge Nodes 1 being children of one or more Router 2 parents. Each Edge Node 1 may have a primary parent Router 2 and, if possible, a secondary parent Router 2 and even additional parent Router(s) 2 for redundancy. Similarly, Routers 2 may have parent-child relationships with other Routers 2. Data Accumulators 3 and/or Gateways 4 may also act as parents.
WSNs in an Automated Meter Reading (AMR) Application
In this discussion, we use as an example a WSN 6 that is configured for automated meter reading (AMR), such as might be provided by a business billing customers for heat, electricity, or water usage. In an AMR application, for example, a Heat Cost Allocator may measure the temperature of a radiator and from that may estimate a customer's energy usage. Similar calculations may be performed with water usage, using a low-powered or self-powered water meter designed for that purpose. The calculation of energy or water usage may be accumulated on the device itself, and a cumulative reading may be reported periodically (such as daily). Readings may be transmitted to a central collection point and then may be sent to an enterprise application. Information may also be displayed locally, such as on an LCD, at the press of a button.
The SensiNet AMR design may be a variant of Sensicast's Distributed Frequency Spread Spectrum (DFSS) design, as described in PCT Application WO05/010214 Method and Apparatus for Wireless Communication in a Mesh Network. For power savings on the Router, a single frequency or small number of frequencies may be used, particularly for control channels used to advertise Router timing and availability.
Although a WSN 6 configured for AMR is used as an example in this discussion, descriptions within are purely illustrative and are not intended to be limiting.
For the AMR application example described herein, we assume a design goal for Routers 2 is to run on battery power, with 10 year battery life at 1000 mAh. A one-day reporting interval may be sufficient, as long as daily reporting is reasonably reliable.
For the AMR application example described herein, we illustrate the design with a hypothetical but typical RF transceiver with the following general characteristics:
The different types of devices in a WSN 6 are discussed in more detail below.
A Router 2 may participate in the formation and operation of the network backbone. It may normally include a sensor and/or actuator and may also relay data from nearby devices. Routers may work together to form a redundant tree topology leading to a Data Accumulator 3. The tree may be redundant; each Router 2 may have multiple parents. Two parents per Router 2 are shown in
A Router 2 may transmit a beacon once each minute on a randomized but predictable schedule. This beacon may allow other nodes to find (e.g., by listening for) the Router 2 and synchronize their timing with the Router 2. When a node becomes the child of a Router 2, it may lock onto the Router 2's beacon schedule, and may read the Router 2's beacon every few minutes to stay synchronized. A node with multiple parents may stay time synchronized with all of its parents. Aspects of the invention related to clock synchronization or other time management techniques are discussed in more detail below.
After each beacon, the Router 2 may wait for a short period of time to allow its children to direct a message toward the Data Accumulator 3. If a Router 2 receives a message from one or more of its children, it may repeat this message to all of its parents. Messages between a Router 2 and its children may be acknowledged. A child may send a message multiple times to its parent, within reason, until the message is acknowledged.
Router 2 memory may be very limited, so only a few messages may be buffered in Router 2 memory. A Router 2 may retain messages from its children only long enough to direct those messages toward the Data Accumulator 3 and/or Gateway 4.
Messages may be sent from a child to a parent during a “contention access period” immediately following the parent's beacon, or at other times that may be scheduled at predictable times in relation to the beacon. The parent/child protocol for a contention access period may be similar to that described in the specification for the IEEE 802.15.4 MAC. With a beaconing interval of about once per minute, the system may have a latency of about one minute per hop.
Routers 2 may also buffer a limited number of messages from a Data Accumulator 3 and/or Gateway 4 to its children, constrained by the limited memory available in the Router 2. A message may be addressed to a specific child device, or it may be addressed to all devices. Indicators of store and forward messages may be included in each beacon, or the message itself may be included in each beacon.
The power draw for a Router 2 is estimated in Table 1. The estimates in Table 1 utilize calculations from Tables 4 and 5 in the Appendix.
Power estimates for Routers.
This assumes that sensor readings are
reported once per day, to all three
parents, with acknowledgements and
Route Child Messages
This power estimate assumes the worst
case scenario, wherein a Router must
relay messages from 500 descendents
once per day.
This estimate assumes a 2 μA power
draw when the device is asleep.
This assumes one 30-byte beacon
transmitted each minute.
This calculation allows a short time
after each beacon to wait for incoming
message from children before going
back to sleep.
This assumes that Routers check for a
message from each of their parents
every two minutes. This also assumes
that Routers simultaneously resyn-
chronize their clock offset to each
parent and that parent-child clocks can
be cross-calibrated with 25 ppm
In order to join the network initially, a
node may run its receiver for a minute
to scan for beacons from all of its
neighbors. This may enable the node
to find and select the best parents.
Beacons may include a measure of link
quality from the Router back to the
Host, thereby enabling devices to pick
the best parents. Routers may rescan
for parents periodically. The power
estimate assumes that a scan is per-
formed every two weeks by each
device for the life of the network.
This estimate corresponds to 11.5 year
battery life from a 1000 mAh battery.
The numbers in Table 1 incorporate very conservative assumptions. In particular, an average of three acknowledged transmissions (i.e., one message and two retries) to each of three parents for each link is assumed.
Sensitivity analysis shows that power consumption as shown in Table 1 may be particularly susceptible to the following variables:
As noted earlier, a WSN 6 may be designed to support hundreds of devices reporting daily. Some Routers 2 may need to expend power passing traffic to and from a large number of more distant devices. For example, Routers 2 at the root of the network (near the Gateway 4) may pass messages to and from nearly every other node in the network, whereas Routers 2 at the edges of the network may need to pass messages to and from only a small number of descendents. Thus, Routers 2 nearer the root of the network may expend more power to service their descendents than may Routers 2 with a smaller number of descendents. Longer contention access periods may be needed for Routers 2 with more descendents. Thus, system capacity may be increased with larger batteries on Routers 2 with a large number of descendents.
Optimizations may be implemented to improve Router 2 battery life. For example, as noted in entry G3 of Table 5 in Appendix A, messages may be combined on heavily used Routers 2, thus decreasing the power required to service children by 50% (for example).
Prior Sensicast specifications such as those described in PCT Application WO05/010214 describe how nodes may synchronize with their parents by listening for messages (called “beacons”) from Routers 2 (or Data Accumulators 3) that have already joined the network. A beacon may be a very short message (on the order of two milliseconds) that may signal the availability of a Router 2 for communication, broadcast a synchronized time base, and/or transmit scheduling information for future beacons so that other nodes may predict the availability of a given Router 2 for communication, so that sleep periods may be synchronized. With precise synchronization, nodes may lessen the amount of time they spend awake listening for beacons, and may thus improve their battery life.
Each Router 2 may transmit a beacon periodically, such as once per minute. New nodes may search for parents by listening for the full beacon interval (e.g., for one minute), and then may track the timing of those parents by listening periodically to subsequent beacons (e.g., every 15 minutes).
The beaconing strategy may be essentially as described in PCT Application WO05/010214. Beacon timing may be randomized to prevent repeated collisions, such as once per minute ±0.5 second (randomized). The randomization schedule may be known and anticipated by the Router's 2 children, based on information included with the beacon, so that children may predict when their parents may be available for communication. The beacon may signal a Router's 2 availability for communication. The beacon may include information about the time base of the WSN 6, so that children may synchronize with their parents. The beacon may also include information about when to expect future beacons, so that child nodes can go to sleep and wake up based on when their parents are next expected to be available for communication.
A Router 2 may select the timing of beacons, as well as the channel on which they are transmitted at a given time, through a combination of user configuration and adaptive algorithms. The timing and channel of beacons may be regular and pseudorandom. For example, a Router 2 may be set to send a beacon every 60 seconds, with a randomized dither of plus or minus 0.500 seconds. The randomized dither may be calculated using a linear congruential generator of the form in Equation 1:
R n+1=(a·R n +b)mod m (1)
The values a (the multiplier), b (the increment), and m (the modulus) are pre-selected constants. The choice of constants is well studied in the computer science literature.
Transmission of the value Rn with each beacon may allow a node to “lock on” to the Router's 2 pseudorandom number sequence. This in turn may be used to forecast the timing of future transmissions, thus allowing the node to wake up and sample the channel at the time a transmission is expected.
Alternatively, and for less computational complexity, the dither may be derived from a lookup table that is shared across the network.
These two techniques may be combined, with a linear congruential generator used to generate a table of a sequence of x pseudorandom numbers. A device wishing to duplicate the table and synchronize with the Router 2 may need the pseudorandom seed used to generate the table, the table length, and the current offset into the table.
For example, a node may use a seed in a linear congruential generator to generate a table of 32 pseudorandom numbers. Each of the table entries may be taken as a dither amount. For example the low-order 10 bits may be used to set the dither in milliseconds, resulting in a dither of ±512 milliseconds (approximately ±0.5 seconds). Thus a Router 2 may send a beacon every minute±a randomized dither from the table. In this example, the true time period to cycle through a table of 32 entries would be 32 minutes±(sum of all 32 randomized dithers).
However, as noted earlier, drift in the clock crystal may require that child devices wake up earlier than the expected beacon time and listen longer, so that they may be sure of hearing a beacon at the appointed time. There may be small timing variations between crystals, and temperature changes may result in additional clock drift. It may be desirable to adjust for these effects, so as to minimize child device wakeup time and thus to prolong battery life. Adjustments for temperature effects and fixed clock skew are discussed below.
Adjusting for Temperature Effects
As noted earlier, power consumption may be fairly sensitive to the degree of time synchronization between the parent and child, which in turn relies on clock accuracy.
Temperature may change the timing characteristics of crystal oscillators. It is possible to buy TCXOs (temperature compensated crystal oscillators) that provide highly accurate timing across a range of temperatures; however, this may involve some cost and may use more power than is desirable. TCXOs may also provide more precision than may be necessary; as noted earlier, the power estimates for the SensiNet AMR design assume 25 ppm accuracy, and TCXOs may achieve much better than 25 ppm accuracy.
A design guide from chip manufacturer Microchip Technology Inc. (PICmicro Microcontroller Oscillator Design Guide AN588, available at http://ww1.microchip.com/downloads/en/AppNotes/00588b.pdf) provides an overview of the issues with various types of timers. The guide describes a formula that estimates how timing may change with temperature in a 32 kHz crystal of the type typically found in an inexpensive WSN 6 device. A formula for temperature-dependent frequency change (in ppm) is given in Equation 2 as:
0.04*((25-° C.)2) (2)
A similar formula may be used by a WSN 6 device to compensate for timing that drifts based on the device's own temperature.
In accordance with one aspect of the invention, a baseline approach for a device with a 32 kHz clock (actually specified as 32.768 kHz in a typical crystal) may be as follows:
For example, the following process may be used to adjust a device's wakeup time period, e.g., the time period between events of beacon transmission or reception:
In a table-driven approach, a mapping between temperature and clock adjustment may be determined empirically for a particular design. Multiple such tables may be used to account for a variety of device profiles. For example, an aged (older or more heavily used) device may have a different set of offsets than a new device, or a device that is experiencing an increase in temperature may have a different profile than a device experiencing a temperature that is stable or decreasing.
Using these techniques, each device on the network may compensate for clock drift as a function of the device's own temperature.
A node's sensor (such as a temperature sensor) may be read and results accumulated every few seconds, in conjunction with clock drift correction. Alternatively, or additionally, a node's sleep cycle may be interrupted by asynchronous sensor events such as the rotation of a water meter.
Adjusting for Fixed Clock Skew
In another aspect of the invention, adjustment may be made for a fixed clock skew that may vary from one device to another, more or less independent of temperature effects. Such adjustment may be performed in conjunction with adjustment for temperature effects, or in the absence of temperature effect compensation. In the SensiNet AMR design, a key purpose of time synchronization may be to ensure that a child node is listening at exactly the time that its parent is scheduled to transmit a beacon. For this limited purpose, it may not be necessary for a device to calibrate the skew of each individual device; rather, the relevant metric may be the sum of the fixed skew for each parent-child relationship, since a device may need to measure only how much its own clock has drifted in relation to its parent's clock drift. This may be measured by child (or parent) devices as described below.
For example, a child node may forecast that one of its parents will beacon in 15 minutes and 3.173 seconds (here noted in the form 15:03.173) in the parent's time base. If the child finds that the beacon actually occurs in 15:03.212 in the child's time base, then the child may calculate a relative clock skew between the parent and child of 0.039 seconds over a 15-minute period. If the next synchronization time is forecast in 15:07.263 in the parent's time base, then the child may apply a fixed offset of 0.039 seconds (or other suitable offset) to forecast a wakeup in 15:07.302 in the child's time base. Such offsets may be averaged, for example to account for the last ten readings. If the fixed clock skew changes with age, this may be accounted for automatically with an approach that is biased in favor of recent data.
Tables 4 and 5 in Appendix A show a calculation to account for Keep-Alive “listening.” This simplified calculation assumes that the maximum uncalibrated clock drift is 25 ppm, and also assumes that a node's receiver is always turned on early enough to hear the beacon. In practice, more adaptive approaches may be used.
Taking the example of the Edge Node 1 Keep-Alive, Table 5 assumes that a node wakes up for a time period of 23.46 ms each 15 minutes. This provides an estimate for making broad design trade-offs, but the actual algorithm that results in this average use of power may work as follows, as shown in the flow chart of
Example of reasonable listening range.
Duration of listening
Offset from forecast
One technique that may be used to calculate the offset may be a formula of the form in Equation 3:
where X may be a value such as 0.25 and where the initial state of OffsetMovingAverage may be 0. With this approach, old values decay exponentially.
Router Messages in SensiNet AMR
In the typical SensiNet WSN 6 implementation, Routers 2 may keep track of their children and may have various buffers reserved for each child. The SensiNet AMR design may not require that Routers 2 know the identities of their children; a Router 2 may simply pass any message from any child toward the Data Accumulator 3. This simplification may dramatically reduce program size and memory requirements.
While the primary flow of messages may be toward a Data Accumulator 3, there may be applications requiring messages (such as NACKs) originating from the Data Accumulator 3 be sent to a specific Router 2 or Edge Node 1. If simplified Routers 2 lack a picture of the rest of the network, other methods are still available for sending messages to a node:
Long addresses, such as 64 bits long, may be assigned to devices at manufacture, but short addresses, such as 16 bits long, may be desirable within the WSN 6 for shorter packets and thus higher power efficiency. When a Data Accumulator 3 first hears from a node with a long ID, it may respond by sending to the node a short ID for use within a particular session.
Forming the SensiNet AMR Network
When a node is powered on, a device may validate that it found the network, such as by flashing an LED. During normal network operation, it may take some time for a Host 5 computer to be notified that a node has joined the network. However, local feedback may be needed quickly, so that an installer may know within a short time, such as within a minute or so, that the node has found the network.
In one implementation, all Routers 2 may be continuously powered. In that scenario, Routers 2 may be installed before the Data Accumulator 3 is installed, and Routers 2 may simply monitor control channels until they hear beacons from Routers 2 that are already connected to a Data Accumulator 3.
Edge Nodes 1 that are battery powered may need to be more selective in their use of power. If they fail to hear beacons when they are first powered, they may go to sleep and may look for beacons only infrequently. The sleep interval may constrain startup time on battery powered devices.
In the SensiNet AMR implementation, Routers 2 may be battery powered, and it may not be realistic to expect that Routers 2 will always be installed at increasing distances from the Data Accumulator 3. This may involve some changes to the method of SensiNet network formation in order to save power. Routers 2 and Edge Nodes 1 may be configured to form clusters of intercommunicating devices, thus sharing a time base and beaconing schedules for low-power operation. When Data Accumulators 3 attach to such clusters, these clusters may, at a future time, be reorganized to direct traffic toward the Data Accumulators 3. In this way, Routers 2 may not need to search continuously for neighbors, and Edge Nodes 1 may attach efficiently to nearby Routers 2 even if a functional data collection network has not yet been formed.
Upon installation, as shown in
As shown in
As shown in
This scheme has the potential to create circular references. A Router 2 may detect circular references by receiving the same message twice. A Router 2 may also detect a circular reference by detecting a path to the Host 5 that includes the Router 2 itself. The Host 5 may also detect circular references by noting when a Router 2 has itself as a descendent. When a Router 2 detects a circular reference, the Router 2 may attempt to rejoin the network.
Similarly, when the Host 5 detects a circular reference, it may instruct the Router 2 to rejoin the network. This may remove the Router 2 from the current path. The Router 2 may then become the child of another Router 2, creating a new path that does not include a circular reference. To prevent the occurrence of circular references, Routers 2 may be instructed to reject paths that contain themselves (i.e., devices may include instructions not to become children of their own children).
An Edge Node 1 may essentially be the same as a Router 2, except that it may not route traffic on behalf of other devices. Therefore, an Edge Node 1 may not need to expend power transmitting beacons or responding to its children. This may allow Edge Nodes 1 to run longer or to utilize smaller batteries.
The power draw for an Edge Node 1 is estimated in Table 3. The estimates in Table 3 utilize calculations from Tables 4 and 5 in the Appendix.
Power estimates for Edge Nodes.
This assumes that sensor readings are reported
once per day, to all three parents, with
acknowledgements and retries.
This estimate assumes a 2 μA power draw
when the device is asleep.
This assumes that Edge Nodes synchronize to
their parents every 15 minutes and
simultaneously resynchronize their
clock offset to each parent. This
also assumes that parent-child clocks
can be cross-calibrated with 25 ppm
This covers overhead to join the network, as
described in Table 1.
This estimate corresponds to 20.0 year battery
life from a 1000 mAh battery.
As with Routers 2, power consumption for Edge Nodes 1 may be particularly sensitive to data rate, clock skew, and sleep current. These factors may be adjusted for as previously described and as shown in
As shown in
System capacity may be limited by the battery capacity of Routers 2 that are a small number of hops away from a Data Accumulator 3 or Gateway 4. Routers 2 with a large number of descendents expend energy transmitting information on behalf of other nodes, and more descendents may result in more traffic. To a certain extent, the system design may compensate by reducing the reporting frequency or combining reports into consolidated packets. But eventually, it will be necessary to transmit more data and/or to extend the Router's 2 contention access period to a point where the battery life goals cannot be achieved. For these situations, it may be necessary to provide larger batteries for Routers 2 that are near the Data Accumulator 3 or Gateway 4.
While it is generally desirable for all Routers 2 to be identical, it may be useful in certain situations to allow certain Routers 2 to be equipped with larger batteries. For example, a network of heat cost allocators and water meters may be installed in an apartment building, with all nodes being configured as Routers 2 for simplicity of installation. These devices may be configured to only accept a certain number of children and descendents. In addition, dedicated Routers 2 with larger batteries may be installed in the hallways as a network backbone, and allowed to establish a working network through the building before any sensors are deployed in the apartments. Due to the routing constraints of the sensor/routers, routes will tend to be directed toward backbone routers with larger batteries.
As noted earlier, Gateways 4 may be mechanically similar to Routers 2, except that, instead of relaying messages, Gateways 4 may provide an interface to a different physical or logical network. Gateways 4 may terminate the WSN 6 protocol and may translate communications to a different protocol appropriate for the new network.
The Gateway 4 may need to support a walk-by scenario, wherein a handheld device may communicate directly with the Data Accumulator 3 to extract the network's data. A different radio may be appropriate for WSN 6 operation vs. a handheld link. For example, in Europe, the 868.0-868.6 MHz band has desirable characteristics for this AMR network design, but devices in this band are limited to a 1% duty cycle due to regulatory constraints. A 1% duty cycle may reduce the available channel from 250 kbps to 2.5 kbps. At 2.5 kbps, data payload throughput (accounting for communications and packet overhead) may be in the range of tens of bytes per second; at that speed, it may take an unacceptably long time to download data for a sizeable network from a Data Accumulator 3 to a handheld device.
A reasonable alternative may be to include a 2.4 GHz IEEE 802.15.4 interface 11 within a Data Accumulator 3 as shown in
For the SensiNet AMR application, the Mini-Gateway 10 may be battery operated, but the battery may be larger than 1000 mAh to provide power for the additional functionality of a Mini-Gateway 10.
In another configuration, the Mini-Gateway 10 may transfer data to another device that may provide Host 5 connectivity through cellular connection or other means. As in other configurations, the Mini-Gateway 10 may collect data, as a Data Accumulator 3 would, and may serve as an interface to a different physical or logical network, as a Gateway 4 would.
The illustrative power model used herein assumes an average of three acknowledged transmissions (the original transmission plus two retries) per link, with three parents per link. Each message may be sent to all three parents for redundancy. If a message does not get through to a Data Accumulator 3 (which may be an infrequent occurrence), the Data Accumulator 3 may notice that the node has not reported, and may send a NACK asking the message to be sent again. If the expected message does not arrive in response to the NACK, the Data Accumulator 3 may send another NACK, within reason. Due to memory constraints on the Routers 2 and other factors, there may be limited bandwidth upstream towards the nodes, but there may be enough capacity to NACK (for example) ten nodes per hour, in addition to periodic messages that are broadcast to all nodes. Ten nodes per hour may cover about 50% of a 500-node network over the course of a day. Thus, the system may provide the following levels of resiliency:
Three parents for each link for redundancy.
For each link, acknowledged communication with retries.
NACK from the Data Accumulator 3 when data is not reported on schedule, essentially providing end-to-end reliability.
Cumulative data reports, so that if one day's data is missed, it may be accumulated in the next day's report.
In this disclosure, we have described how data may be reported from device to device within a WSN. We have considered how data from a network of devices may be accumulated and passed to devices or applications outside the network. We have discussed some requirements for timely, low power communication within such WSNs, including methods of clock synchronization and network backbone formation. We have also addressed features that may be incorporated to improve system reliability. Again, the embodiments described herein are meant to be illustrative and are not intended as limiting. In addition, various features described above may be combined in any suitable way to form a system in accordance with the invention.
Power calculations for WSN.
Contention access period
Listen for children
Periodically search for
Router battery capacity
Edge node battery capacity
Router keep alive interval
Check for parent
Edge node keep alive interval
Synchronize clock to
Number of parents
tertiary, etc . . .
Average number of
retries per hop
Edge Node Power Draw per Annum (see Table 5 for detailed calculations)
Table 5 cell A10
Table 5 cell B2
Table 5 cell D8
Table 5 cell F4
Router Power Draw per Annum (see Table 5 for detailed calculations)
Table 5 cell A10
Table 5 cell B2
Table 5 cell E8
Table 5 cell G4
Table 5 cell C11
Table 5 cell F4
Detailed power calculations for WSN.
Comments and Formulas (using cell numbers on left)
(warmup_time & warmup_current taken from Table 4)
Edge Node Report
Transmit per report
(A2 * 8)/C1
Copied from transmit time A3
Number of parents
(A7 * A8) * ((warmup_time * warmup_current) +
(A3 * A4) + (A5 * A6))/(1000 * 60 * 60 * A1)
Draw per annum
Draw per annum
B1 * 24 * 365/1000
Router Periodic Beacon
Time per beacon
Average draw transmit
((warmup_time * warmup_current) + (C3 * C4))/
(C5 * 60 * 1000)
(C7 * 8)/C1
Average draw receive
(C8 * C9)/(C5 * 60 * 1000)
Draw per annum
(C6 + C10) * 24 * 365
Edge Node Keep-Alive
D2 * 60 * 1000/10000000
Accounting for crystal drift; C3 + (D1 * D3)
((warmup_time * warmup_current) + (D4 * D5))/
(D1 * 60 * 1000)
Parents to track
Draw per annum
D6 * D7 * 24 * 365
E2 * 60 * 1000/10000000
Accounting for crystal drift; C3 + (E1 * E3)
((warmup_time * warmup_current) + (E4 * E5))/
(E1 * 60 * 1000)
Parents to track
Draw per annum
E6 * E7 * 24 * 365
Router beacon interval
Rescynchs per annum
Power per annum
F1 * F2 * F3/60
Number of descendents
Power draw for each
Rough estimate -- twice Edge Node report
(inbound and outbound); 2 * A10
Combine messages on heavily used Routers
Power per annum
G1 * G2 * G3
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US5694335 *||Mar 12, 1996||Dec 2, 1997||Hollenberg; Dennis D.||Secure personal applications network|
|US6326825 *||Jan 18, 2001||Dec 4, 2001||Agilent Technologies, Inc.||Accurate time delay system and method utilizing an inaccurate oscillator|
|US6453181 *||Nov 4, 1999||Sep 17, 2002||Qualcomm, Incorporated||Method and apparatus for compensating for frequency drift in a low frequency sleep clock within a mobile station operating in a slotted paging mode|
|US6844814 *||Nov 12, 2002||Jan 18, 2005||Motorola, Inc.||Wireless sensor apparatus and method|
|US7096359 *||Mar 1, 2001||Aug 22, 2006||University Of Cincinnati||Authentication scheme for ad hoc and sensor wireless networks|
|US7119676 *||Jul 22, 2004||Oct 10, 2006||Innovative Wireless Technologies, Inc.||Method and apparatus for multi-waveform wireless sensor network|
|US7135976 *||Mar 29, 2004||Nov 14, 2006||Rftrax, Inc.||Wireless monitoring device|
|US7242294 *||Sep 17, 2003||Jul 10, 2007||Agilent Technologies, Inc||System and method for using mobile collectors for accessing a wireless sensor network|
|US7313399 *||Jun 4, 2004||Dec 25, 2007||Millennial Net, Inc.||Protocol for configuring a wireless network|
|US7324824 *||Dec 8, 2004||Jan 29, 2008||Awarepoint Corporation||Wireless network monitoring system|
|US7356561 *||May 1, 2003||Apr 8, 2008||Lucent Technologies Inc.||Adaptive sleeping and awakening protocol for an energy-efficient adhoc network|
|US7593117 *||Dec 14, 2006||Sep 22, 2009||The Boeing Company||Apparatus and methods for measuring workpieces|
|US7675863 *||Jul 12, 2007||Mar 9, 2010||Sensicast Systems||Method and apparatus for wireless communication in a mesh network using connectivity assessment at multiple frequencies|
|US7701858 *||Apr 30, 2004||Apr 20, 2010||Sensicast Systems||Method and apparatus for wireless communication in a mesh network|
|US20010028301 *||May 31, 2001||Oct 11, 2001||Klever Marketing, Inc.||Electronic shopping cart display system|
|US20030151513 *||Nov 21, 2002||Aug 14, 2003||Falk Herrmann||Self-organizing hierarchical wireless network for surveillance and control|
|US20030152041||Nov 21, 2002||Aug 14, 2003||Falk Herrmann||Protocol for reliable, self-organizing, low-power wireless network for security and building automation systems|
|US20040090329 *||Oct 24, 2003||May 13, 2004||Hitt Dale K.||RF based positioning and intrusion detection using a wireless sensor network|
|US20040100917||Nov 26, 2002||May 27, 2004||Perkins Matthew R.||Coordinator device election process for wireless sensor networks|
|US20040218602 *||Feb 6, 2004||Nov 4, 2004||Hrastar Scott E.||Systems and methods for dynamic sensor discovery and selection|
|US20040235468 *||May 19, 2003||Nov 25, 2004||Luebke Charles J.||Wireless network clustering communication system, wireless communication network, and access port for same|
|US20050093702 *||Nov 12, 2004||May 5, 2005||Twitchell Robert W.Jr.||Manufacture of LPRF device wake up using wireless tag|
|US20050155839 *||May 13, 2004||Jul 21, 2005||J.J. Mackay Canada Limited||Efficient battery powered electronic parking meter|
|US20050200480 *||Dec 8, 2004||Sep 15, 2005||Caras Jay D.||System and method for monitoring the occurrence of situational and environmental events using distributed sensors|
|US20060002368 *||Jul 1, 2004||Jan 5, 2006||Honeywell International Inc.||Latency controlled redundant routing|
|US20060029061 *||Oct 6, 2004||Feb 9, 2006||Dust Networks||Low-power autonomous node for mesh communication network|
|US20060046664 *||Aug 26, 2004||Mar 2, 2006||Massachusetts Institute Of Technology||Parasitic mobility in dynamically distributed sensor networks|
|US20070115827 *||Dec 17, 2004||May 24, 2007||Sony Deutschland Gmbh||Wireless communication network architecture|
|US20070258508 *||Apr 30, 2004||Nov 8, 2007||Werb Jay P||Method and apparatus for wireless communication in a mesh network|
|US20070276270 *||May 24, 2006||Nov 29, 2007||Bao Tran||Mesh network stroke monitoring appliance|
|US20080004904 *||Aug 30, 2006||Jan 3, 2008||Tran Bao Q||Systems and methods for providing interoperability among healthcare devices|
|US20080036589 *||Jul 12, 2007||Feb 14, 2008||Sensicast Systems||Method and apparatus for wireless communication in a mesh network using connectivity assessment at multiple frequencies|
|US20080037431 *||Jul 12, 2007||Feb 14, 2008||Sensicast Systems||Method and apparatus for wireless communication in a mesh network with trackable randomized schedule|
|US20080037454 *||Jul 12, 2007||Feb 14, 2008||Sensicast Systems||Method and apparatus for wireless communication in a mesh network with software downloaded to nodes|
|US20080037569 *||Jul 12, 2007||Feb 14, 2008||Sensicast Systems||Method and apparatus for wireless communication in a mesh network using software proxies|
|US20080040509 *||Jul 12, 2007||Feb 14, 2008||Sensicast Systems||Method and apparatus for wireless communication in a mesh network with node activity monitoring|
|US20080273547 *||May 1, 2007||Nov 6, 2008||Honeywell International, Inc.||Apparatus and method for acknowledging successful transmissions in a wireless communication system|
|EP1661318A2 *||Jul 16, 2004||May 31, 2006||Sensicast Systems, Inc.||Method and apparatus for communication in a mesh-type wireless sensor network|
|WO2005010214A2 *||Jul 16, 2004||Feb 3, 2005||Sensicast Systems Inc||Method and apparatus for communication in a mesh-type wireless sensor network|
|1||Annex dated May 5, 2006, International Application No. PCT/US2005/046169.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US8249049 *||Mar 17, 2009||Aug 21, 2012||Cisco Technology, Inc.||Clock synchronization|
|US8537721 *||Aug 7, 2009||Sep 17, 2013||Koninklijke Philips N.V.||Method for scheduling transmissions of global beacons in body area networks|
|US9251253||Jan 5, 2013||Feb 2, 2016||Qualcomm Incorporated||Expeditious citation indexing|
|US20100238917 *||Mar 17, 2009||Sep 23, 2010||Silverman Matthew Aaron||Clock synchronization|
|US20110176459 *||Aug 7, 2009||Jul 21, 2011||Koninklijke Philips Electronics, N.V.||Method for scheduling transmissions of global beacons in body area networks|
|US20130286909 *||Apr 26, 2012||Oct 31, 2013||Qualcomm Atheros, Inc.||System and method for reducing power consumption in a wireless communication system|
|U.S. Classification||370/311, 370/252, 455/574, 370/238, 370/254, 370/436, 370/389, 370/338, 370/310, 455/343.1, 370/347, 455/572|
|International Classification||H04B1/16, H04L12/66, G08C17/00|
|Cooperative Classification||H04L67/12, Y02B90/246, Y02B90/242, H04W40/32, Y04S20/42, H04W52/0216, G01D4/004, H04W84/18, Y04S20/322|
|European Classification||H04L29/08N11, G01D4/00R1, H04W52/02T2A|
|Apr 27, 2006||AS||Assignment|
Owner name: SENSICAST SYSTEMS, MASSACHUSETTS
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:WERB, JAY PHILIP;BERRY, VICTOR A.;REEL/FRAME:017550/0356
Effective date: 20060404
|Nov 16, 2011||AS||Assignment|
Owner name: SENSICAST SYSTEMS, INC., MASSACHUSETTS
Free format text: CORRECTION BY DECLARATION OF ASSIGNEE NAME AND STATE OF INCORPORATION RECORDED AT REEL/FRAME 17550/0356;ASSIGNOR:SENSICAST SYSTEMS, INC.;REEL/FRAME:027241/0494
Effective date: 20111116
|Apr 4, 2012||AS||Assignment|
Owner name: SENSICAST SYSTEMS, INC., MASSACHUSETTS
Free format text: CORRECTIVE ASSIGNMENT TO CORRECT THE ASSIGNOR NAME PREVIOUSLY RECORDED ON REEL 017550 FRAME 0356. ASSIGNOR(S) HEREBY CONFIRMS THE CORRECTION BY DECLARATION OF ASSIGNEE NAME AND STATE OF INCORPORATION RECORDED AT REEL/FRAME 17550/0356;ASSIGNOR:SENSICAST SYSTEMS;REEL/FRAME:027987/0670
Effective date: 20111116
|Jan 30, 2013||AS||Assignment|
Owner name: HONEYWELL INTERNATIONAL SARL, SWITZERLAND
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:SENSICAST SYSTEMS, INC.;REEL/FRAME:029726/0029
Effective date: 20120419
Owner name: YOKOGAWA USA, INC., GEORGIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:SENSICAST SYSTEMS, INC.;REEL/FRAME:029726/0029
Effective date: 20120419
|Feb 25, 2015||FPAY||Fee payment|
Year of fee payment: 4
|Sep 4, 2015||FPAY||Fee payment|
Year of fee payment: 4
|Sep 4, 2015||SULP||Surcharge for late payment|